For all patients, the efficacy of new treatments for AD and other dementias may be enhanced when administered during the earliest diagnosable stage of disease. This requires efficient strategies for screening and diagnosing very early dementia at high levels of sensitivity, specificity, and reliability. To achieve this goal in primary care settings, test performance should be minimally influenced by race/ethnicity and education and more sensitive to the memory impairment associated with very early dementia. Our long-term objective is to design efficient culture-fair strategies to identify patients with very early dementia at cross-section for treatment in primary care or for participation in clinical trials. We hypothesize that this will involve rapid screening of all geriatric patients and limited additional testing in patients who screen positive in order to demonstrate memory impairment and another cognitive disturbance as required by DSM IV criteria. A clinical evaluation is triggered only for patients who meet neuropsychological criteria for dementia, making screening and diagnosis feasible in primary care settings. We will also examine the influence of race/ethnicity and education on the diagnostic process. 250 African American and 250 Caucasian patients and their informants will be recruited from an urban geriatric medicine clinic and randomly assigned to 1 of 4 diagnostic conditions defined by whether or not raters know the patient's race and/or education. Diagnosis and clinical staging will be established by consensus at each of 3 waves independent of screening tests and results from other waves. A consensus gold standard diagnosis will be established at the end using the patient's longitudinal record. From the results of this study, we will recommend strategies to maximize the identification of patients with very early dementia at cross section. The specific aims include: Specific Aim 1: To estimate the sensitivity and specificity of each test and test combination for identifying prevalent dementia. Specific Aim 2: To estimate the specificity of each test and test combination at fixed levels of sensitivity for identifying incident dementia. Specific Aim 3: To estimate the effect of knowing race and education on the diagnostic process.